Instructions to use ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth")
model = AutoModelForMultimodalLM.from_pretrained("ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth

SGLang

How to use ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth with Docker Model Runner:
```
docker model run hf.co/ThakiCloud/Qwen3.8-27B-Satoori-KO-Synth
```

thaki-AI commited on 6 days ago

Commit

7f33762

verified ·

1 Parent(s): ea53dd0

Add SK5/SK12-D9b LoRA adapters + merge code

Browse files

Files changed (1) hide show

adapters/coupled-8600/README.md +35 -0

adapters/coupled-8600/README.md ADDED Viewed

	@@ -0,0 +1,35 @@

+# SK5 — coupled arm (8,600 examples)
+LoRA adapter (r=8, alpha=16, all-linear, lr 2e-5, one epoch) for `Qwen/Qwen3.8-27B`.
+The **coupled** half of the primary comparison in *Metric--Construction Coupling Inflates
+Measured Synthetic Dialect Recovery*. Its transformation rules used the **full** marker
+inventory --- the same inventory the marker metric scores against.
+Pair with [`adapters/disjoint-8600`](../disjoint-8600), which differs only in that its rules
+used 80% of that inventory, to reproduce the paper's Figure 3.
+## ⛔ 반드시 병합한 뒤 서빙하세요
+**어댑터를 그대로 붙여 서빙하면 이 아키텍처에서는 조용한 no-op 이 됩니다.** 오류도 경고도
+없이 base 모델과 같은 출력이 나옵니다(자체 실측). 반드시 아래처럼 병합한 가중치로 서빙하세요.
+```bash
+# merge_lora.py 는 이 저장소에 함께 있습니다 (496/496 모듈 적용 + 샤드 검증 내장)
+python merge_lora.py \
+  --base  Qwen/Qwen3.8-27B \
+  --adapter adapters/coupled-8600 \
+  --out   /path/to/merged
+# 출력 마지막 줄이 MERGE_OK {"n_applied": 496, "verify": "PASS"} 인지 확인하세요
+```
+## 출처 (AI 허브)
+아래 데이터로 **직접 학습**해 만든 가중치입니다. 제공처 약관상 학습으로 생성된 모델·가중치는
+2차 저작물로 공개할 수 있으나, **원본 데이터 및 원본의 단순 재가공 데이터는 포함하지 않습니다.**
+- 한국어 방언 발화 (강원 118 · 경상 119 · 전라 120 · 제주 121 · 충청 122)
+- 중·노년층 한국어 방언 데이터 (71517, 71558)
+- 출처: AI 허브 (https://aihub.or.kr)
+⛔ 이 폴더에는 원본 문장·방언형 대응 목록·마커 어휘집이 **들어 있지 않습니다.**